Slopwatch: The Typical Slopfarms and the 'Brian Fagioli Dilemma'
Last week we saw some articles about IPFire 2.29 Core Update 193 [1, 2] and grouped them. Days later the slopfarm called linuxsecurity.com published this:
It's LLM slop, as usual.
So was the other fake 'article' (from the same day):
As almost 100% of their output is LLM slop this did not surprise us:
Sometimes they add some stock image (like project logo), but that does not change the fact we're dealing with LLM slop. We noticed a number of other sites (as lately as yesterday) that claim to cover "Linux" and instead use LLM slop regarding "Linux". We won't name them yet, we're just keeping them under observation for now (maybe they only experiment a little and will refrain from repetition).
There's a real risk that not only will such sites realise LLMs don't pay off; when they revert back to real writing nobody will believe them anymore. That is what happened to Brian Fagioli. He misused LLM slop to the point where even if he actually writes something, the default assumption is that it's fake. His editor had to intervene a number of times; he even deleted some Fagioli articles that were clearly slop.
Looking at Google News right now, a few hours it promoted this LLM slop:
It's a fake 'article' (even the image is slop!) and, as usual, it is Linux-hostile, probably generated by a Microsoft LLM.
To the Web and to society (exposed to the Web) LLMs are a net negative. After Microsoft filled E-mail with SPAM it does the same to the Net at large. It just doesn't care. █